FareedKhan-dev/KG-Pipeline ? reverse-engineered prompt

Build me a Jupyter notebook that turns a chunk of plain text into an interactive knowledge graph, step by step, so it is easy for a beginner to follow. I want it to take unstructured text, send it to an LLM through an OpenAI compatible API, extract subject, predicate, object facts, clean up duplicates, then build a graph from those facts and show it right inside the notebook. Please include clear cells for setup, importing libraries, entering the text, splitting it into chunks if needed, prompting the model, collecting the extracted triples, normalizing them, and then visualizing the final graph interactively. Use environment variables for the API key and make it easy to switch models. If you need to look up current docs online for the notebook widgets or API details, go ahead. Keep the code readable and beginner friendly, with short explanations between steps and visible outputs at each stage.